This repository contains the documentation about the data model and development process of IICONGRAPH.
Python 3.8.10 must be installed in your machine. Either Blazegraph or GraphDB must also be installed. Jupyter notebook must also be installed (either normal version or JupyterLab). Please refer to the corresponding documentations to install these softwares/programming language.
You can install the requirements by running pip install -r requirements.txt
IICONGRAPH mainly uses the ICON ontology as its schema. ICON is an ontology developed to describe artistic interpretation according to the metholodology of Panofsky. According to Panofsky, an interpretation can be divided into three communicating levels. On the first level, named preiconographical, you associate actions, expressional qualities and natural elements (like a tree, dog, sword) with artworks by recognizing them as artistic motifs. Meanwhile the natural elements, expressional qualities and actions are universal, the artistic motifs are the representation of those universal concepts in a painting. I.e., you state that there is an horse (concept) in Guernica (artwork) because you recognize that there is an artistic motif which resembles the concept of horse. In other words, two different painting may share the same natural element but they do not share the artistic motifs which represent it. You can also recognized sets of artistic motifs (artistic motif of woman + artistic motif of sadness) together in a composition. On the second level, named iconographical, you associate Characters, Events, Named Objects (Thor's Hammer), Symbols, Personifications, Places (Rome, if you recognize a general "city" that would be a natural element) with an artwork by the recognition of images. Images work in the same way as artistic motif, the "Character" is universal, the "image" is the representation of that character in that specific artwork. When you do an iconographical recognition, you can also refer to a previous pre-iconographical recognition. (i.e, you recognize that the Image of Athena in this painting is associated with the artistic motif of a woman that you recognize in a pre-iconographical level. This is important because you could recognize Zeus as a character and then refer to the artistic motif of a Bull as Zeus is sometimes depicted as a Bull). Symbols are expressed by following the Simulation Ontology, as in HyperReal (described below). In this ontology, a Simulation is the link between a symbol (also called Simulacrum), its symbolic meaning (also called Reality Counterpart) and the cultural context in which the symbol symbolizes the symbolic meaning. In this level you can also recognize stories and allegories, which are sets of all the iconographical elements named above (Characters, events, etc...). On the third level, named iconological, you associate abstract concepts or cultural phenomenon in artworks via the recognition of its intrinsic meaning. The intrinsic meaning can be connected to previously recognized images and/or artistic motifs to support the claim.
Table 1 below summarizes the ICON and Simulation Ontology classes (plus classes from other ontologies they reuse) used in IICONGRAPH and Table 2 summerizes the object properties used (from ICON, Simulation Ontology, and other ontologies that are reused in them).
| Class | Description | Notes |
|---|---|---|
| https://w3id.org/icon/ontology/Artwork | This class represents the artworks in the Graph | |
| https://w3id.org/simulation/ontology/Simulation | This class represents the normal simulations in the graph | |
| https://w3id.org/icon/ontology/Symbol | This class represents the symbols in the graph (it is aligned to Simulation) | |
| https://w3id.org/simulation/ontology/AttributeSimulation | This class represents the simulations in which the symbolic meaning is a symbolic attribute of the symbol (i.e., "Cornucopia" as the attribute of the personification of Abundance) | Subclass of Simulation |
| https://w3id.org/simulation/ontology/EmblematicSimulation | This class represents a simulation in which the symbol meaning is an emblem of the symbol | Subclass of Simulation |
| https://w3id.org/simulation/ontology/AssociationSimulation | This class represents a simulation where the symbolic meaning and the symbol are associated (therefore it is reflexive) | Subclass of Simulation |
| https://w3id.org/simulation/ontology/CorrespondenceSimulation | This class represents the simulations in which the symbolic meanings symbolically corresponds to the symbol | Subclass of Simulation |
| https://w3id.org/simulation/ontology/ProtectionSimulation | This class represents the simulations in which the symbolic meaning is symbolically prevented by the symbol | Subclass of Simulation |
| https://w3id.org/simulation/ontology/ManifestationSimulation | This class represents the simulations in which the symbolic meaning symbolically manifests the symbol | Subclass of Simulation |
| https://w3id.org/simulation/ontology/RelatednessSimulation | This class represents the simulations in which symbols and symbolic meanings are generally related. | Subclass of Simulation |
| https://w3id.org/simulation/ontology/HealingSimulation | This class represents the simulations in which the symbol is a symbolic cure for the symbolic meaning. I.e., liquorice as a symbolic cure for envy. | Subclass of Simulation |
| https://w3id.org/icon/ontology/IconographicalRecognition | A Iconographical Recognition is a n-ary relationship class that links the artwork described (and, if available, the Artistic Motifs recognized at the previous level along with their compositions) to the Images recognized, or directly to the Invenzione that results from the grouping of the subject depicted by each single image. The N-ary structure is necessary in order to include relevant additional information concerning the visual interpretation claimed. | |
| http://www.ontologydesignpatterns.org/ont/dul/DUL.owl#Agent | The agent represents the interpreter, and can be either human or a machine (in the case of the automatic interpretations) | |
| https://w3id.org/icon/ontology/Image | Image is a reification class linking the identified second level subject (namely: Characters, Events, Places, Named Objects, Symbols and Personifications) to the specific interpretation that recognizes it. It represents the subject depicted as a manifestation in the specific artwork taken into account, in contrast to the second level subjects cited and Invenzione, identifying iconographies from an abstract and general point of view. This distinction between the general subject level and the artwork-specific one is functional to identify the variants of a subject in relation to the specific context. | |
| https://w3id.org/icon/ontology/ArtisticMotif | The artistic motif is the representation of a pre-iconographical element in an artwork | |
| https://w3id.org/icon/ontology/Composition | A composition is a set of artistic motifs | |
| https://w3id.org/icon/ontology/Event | An Event is an action that is depicted as subject by an artwork. It differs from the first level Action since it requires deeper interpretation to be identified. For example, the act of hurling the Trident (Action) held by Neptune in some iconographies is interpreted as the act of calming the storm (Event) provoked by Juno in Virgil's Aeneid book 1. Furthermore, an Event is linked to a icon:Image, and not to a icon:Artistic Motif. This class includes also events having a proper identity, such as historical events (e.g. the Battle of Cascina). | |
| https://w3id.org/icon/ontology/PreiconographicalRecognition | A Pre-iconographical recognition is a n-ary relationship class that links the artwork described (and optionally the Pure Forms composing it) to the Artistic Motifs recognized as carrier of meaning at the first level of analysis. The N-ary structure is necessary in order to include relevant additional information concerning the visual interpretation claimed. | |
| https://w3id.org/icon/ontology/Character | A Character is a real or imaginary being identifiable as a subject in the Iconographical Analysis (level 2 of interpretation) | |
| https://w3id.org/icon/ontology/Place | An event is an iconographical element that can be represented via an image in an artwork. | |
| https://w3id.org/icon/ontology/NaturalElement | Comprises objects, human being, natural elements that can be experienced. They are recognized at the first level of interpretation. | |
| https://w3id.org/icon/ontology/Story | A story is a cohese combination of Characters, Places, Events, Named Objects. The recognition of a story happens in the context of an Iconographical Recognition, and it is included in an Iconographical Description. As the super class Invenzione, it concerns the word of the subject matter. It is not necessary to have multiple iconographic elements to compose a story. | |
| http://www.ontologydesignpatterns.org/ont/dul/DUL.owl#Quality | The poses that the characters depicted may have are here considered as Qualities. The compositional principles concerning the human body representation, such as the contrapposto, are considered as Qualities as well, since they can be interpreted as poses assumed by the human figure. Every specification of the element (e.g. NaturalElement, Action) described, such as color, position, age, may be described as quality. | |
| https://w3id.org/icon/ontology/Action | An event as recognized at level 1 of interpretation, e.g. a man who lifts one hand or a man riding a horse. Generally, all the actions should be described by this class. When an action is identified, it is suggested to create a Composition with all the elements involved in this action | |
| https://w3id.org/icon/ontology/ExpressionalQuality | An Expression Quality corresponds to the "expressional subject matter" described by Panofsky (1939), which belongs, along with the factual ones, to the Primary or natural subject matter of a work of art. Indeed, it individuates, by way of example, emotions, an atmosphere, the character of a person. | |
| https://w3id.org/icon/ontology/NamedObject | A named object is a non-living unique element that is often used as an attribute for the recognition of specific characters or as the individuals of a more general concept class. (e.g. Bible for the Book class) |
Wikidata is a collaborative knowledge base containing information about a wide range of domains. We are interested in the painting domain, and we converted the statements that linked paintings to the elements they depict.
ArCo is the Italian cultural heritage knowledge graph. We are interested in all the Historic or Artistic properties that have a textual description following the "Iconographic Reading" syntax (such as this).
The description is in Italian, and, as seen in the scripts, an automatic translation is necessary before the conversion.
HyperReal a knowledge graph of cultural symbolism. It is used in the context of IICONGRAPH to generate automatic potential symbolic interpretation. Imagine a painting depicts an apple, and an apple in HyperReal is the symbol of the original sin from a Christian point of view. The painting depicting an apple can be automatically interpreted as symbolizing the original sin. These interpretations are creator-agnostic and just potential, and should be used as a starting point for art historian to analyse the symbolism behind artworks.
For the data conversion of Wikidata, please refer to the jupyter notebook IICONGRAPHwikidatadocs.ipynb
For the data conversion of ArCo, please refer to the jupyter notebook IICONGRAPHarcodocs.ipynb
The released version of IICONGRAPH is the combined version created by exporting function of GraphDB after loading both IICONGRAPHarco and IICONGRAPHwikidata in the same instance.
The same result can be obtained in Python by using the library rdflib in the following code
from rdflib import Graph
g1 = Graph()
g2 = Graph()
g1.parse("path_to_iicongrapharco")
g2.parse("path_to_iicongraphwikidata")
combined_graph = Graph()
combined_graph = g1 + g2
combined_graph.serialize(destination="namefile.ttl", format="ttl")
# Considering the size of the graphs, it might take a while for rdflib
#to combine themThe table below contains quantitative data about the content of IICONGRAPHglobal and its two subgraphs (IICONGRAPHwiki and IICONGRAPHarco).
| Number of… | IICONGRAPHarco | IICONGRAPHwiki | IICONGRAPHglobal |
|---|---|---|---|
| triples | 767888 | 29148273 | 29900414 |
| preiconographical subjects | 110362 | 224981 | 335343 |
| unique preiconographical subjects | 11651 | 5131 | 16782 |
| iconographical subjects | 10500 | 121893 | 132393 |
| unique iconographical subjects | 3641 | 37667 | 41308 |
| inferred interpretations | 457747 | 3030142 | 5128860 |
| uniquely inferred interpretations | 8030 | 13131 | 16064 |
| iconological subjects | 112956 | 0 | 112956 |
| unique iconological subjects | 5610 | 0 | 5610 |
The catalogue data of IICONGRAPH1.0 is released in the turtle serialization, following the DCAT model it is available here
- [1]Bruno Sartini, Sofia Baroncini, Marieke van Erp, Francesca Tomasi, and Aldo Gangemi. 2023. ICON: An Ontology for Comprehensive Artistic Interpretations. J. Comput. Cult. Herit. 16, 3, Article 59 (September 2023), 38 pages. https://doi.org/10.1145/3594724
- [2] Panofsky, E.: Studies in iconology: humanistic themes in the art of the Renaissance. Westview Press, Boulder, Colo. (1972)
- [3] Bruno Sartini, Marieke van Erp, and Aldo Gangemi. 2021. Marriage is a Peach and a Chalice: Modelling Cultural Symbolism on the Semantic Web. In Proceedings of the 11th Knowledge Capture Conference (K-CAP '21). Association for Computing Machinery, New York, NY, USA, 201–208. https://doi.org/10.1145/3460210.3493552
- [4] Vrandeˇci ́c, D., Kr ̈otzsch, M.: Wikidata: A free collaborative knowledgebase. Commun. ACM 57(10), 78–85 (sep 2014). https://doi.org/10.1145/2629489
- [5] Carriero, V.A., Gangemi, A., Mancinelli, M.L., Marinucci, L., Nuzzolese, A.G., Presutti, V., Veninata, C.: Arco: The italian cultural heritage knowledge graph. In: Ghidini, C., Hartig, O., Maleshkova, M., Sv ́atek, V., Cruz, I., Hogan, A., Song, J., Lefran ̧cois, M., Gandon, F. (eds.) The Semantic Web – ISWC 2019. pp. 36–52. Springer International Publishing, Cham (2019)